CRAZY ILLUSIONS can be created by the power of gravity. Objects

Transcription

CRAZY ILLUSIONS can be created by the power of gravity. Objects
CRAZY ILLUSIONS can be created by the power of gravity.
Objects can be multiplied manyfold—as in this case of
a certain magazine’s logo acted on by a computer program
that simulates the effect of gravity on light. Or they can
be magnified and mangled—like the galaxies distorted by the
galaxy cluster Abell 2218 (opposite page). The large yellowish
galaxies belong to the cluster; the thin bluish curves are the
images of galaxies five to 10 times farther away from us.
Copyright 2001 Scientific American, Inc.
The most massive telescopes
known to humanity sit not on earthly
mountaintops but in deep space.
They are gravitational lenses, once
mere curiosities, now one of the
most important tools in astronomy
By Joachim Wambsganss
Gravity’s
MARK CLEMENS, USING ADOBE PHOTOSHOP; GRAVITATIONAL LENS FILTER BY PETE KERNAN
Case Western Reserve University (theory2.phys.cwru.edu/~ pete) (opposite);
ANDREW FRUCHTER Space Telescope Science Institute and NASA (Abell 2218)
Kaleidoscope
T
o many people, the universe seems like a hall of mirrors— filled
with objects that are beyond bizarre and phenomena that challenge our very understanding of reality. Little do they realize how
apt this metaphor is. The skies are riddled with fun-house illusions: quasars that appear in quadruplicate, galaxies that are
squished from their usual pinwheel or beehive shape into long,
skinny threads, stars that fade in and out like streetlamps on a
foggy night. Just as psychologists prize optical illusions for what
they reveal about the brain, astronomers find that the heavenly
mirages show a universe they might not otherwise see.
Usually light from an astronomical object goes straight from
the object through the depths of space into our telescopes. But
if a second object is located exactly in between, its gravity can
deflect the light, much as a glass lens does. We see a distorted,
magnified or multiplied image. Analysis of that image can shed
light both on the background object and on the lens itself.
The study of gravitational lensing is still a young field, having just finished its teenage years as an observational science. A
decade or so ago astronomers knew of just a few examples of
lenses [see “Gravitational Lenses,” by Edwin L. Turner; Scien-
tific American, July 1988]. They have since detected and explored entire new manifestations of lensing: the so-called microlensing of quasars and stars; arclets and weak lensing in
galaxy clusters; and, last year, the subtle shearing caused by very
weak lensing of the large-scale structure of the universe. Anything that possesses mass can serve as a lens; it does not need to
emit light of its own. For this reason, gravitational lensing is one
of the few ways that astronomers can map out the invisible dark
matter of the universe. Lensing can also probe the internal structure of quasars, spot black holes traipsing through interstellar
space and detect Earth-mass planets around other stars.
Credit for gravitational lensing is often given to Albert Einstein, but he was not, in fact, the first person to predict it. As early as 1801, Berlin astronomer and geographer Johann Georg
von Soldner argued that the attractive force of the sun could
www.sciam.com
SCIENTIFIC AMERICAN
Copyright 2001 Scientific American, Inc.
65
THE AUTHOR
bend the light rays of distant stars. According to Newtonian
gravity theory, the position of a star seen near the edge of the
sun should shift by 0.84 arcsecond relative to its position measured half a year later, when the sun is elsewhere in the sky.
According to general relativity, however, the angle is twice
as large. As Einstein wrote, “Half of this deflection is produced
by the Newtonian field of attraction of the sun, and the other
half by the geometrical modification (‘curvature’) of space
caused by the sun.” During the now famous solar eclipse of
May 1919, British astrophysicists Arthur S. Eddington and
Frank W. Dyson measured this effect and confirmed the relativistic estimate (although, in retrospect, the experimental precision was probably insufficient to distinguish the two estimates
beyond a reasonable doubt).
Einstein dealt again with gravitational light deflection in the
1930s, when he predicted that a foreground star could magnify the image of a background star. But he was skeptical that such
an illusion could ever be seen. More optimistic were Swiss-
JOACHIM WAMBSGANSS often compares gravitational lensing to
looking through the bottom of a wineglass. Both produce the same
type of distortion. But what makes the comparison so apt is that
Wambsganss comes from a family of winemakers; his uncles,
cousins and parents own vineyards in the Rhine Valley. He says
he became interested in cosmology when a high school teacher
gave him a copy of Scientific American. (No, we didn’t bribe him to
say that.) Today he is a physics professor at Potsdam University.
American astrophysicist Fritz Zwicky, who predicted the lensing effects of galaxies and galaxy clusters, and American Henry Norris Russell, who suggested that this light deflection could
be used to visualize and popularize relativity [see “A Relativistic Eclipse,” by Henry Norris Russell; Scientific American,
February 1937]. It was not until 1979, however, that astronomers actually saw evidence of lensing. The following pages
review the progress since then.
FOUR CONSEQUENCES OF GRAVITATIONAL LIGHT DEFLECTION
CHANGE OF POSITION The deflection
shifts the apparent location of a star,
galaxy or quasar in the sky. In most
cases, this makes little difference to
observers, because they do not know
where the object would have been in the
absence of lensing. But if the sourcelens alignment changes—for instance, if
either is moving— astronomers can directly measure the displacement.
■
■
■
66
LENSES
MAGNIFICATION
SOURCE
RESULT
MAGNIFICATION AND DEMAGNIFICATION
The deflection and focusing of light rays
affect the apparent brightness of the
background star or quasar. Although
most cosmic sources are demagnified
slightly, some are magnified by varying
degrees. Observers have measured magnifications of more than 100 times.
DEFORMATION Extended cosmic objects
(such as galaxies) often appear stretched
along a circle centered on the lens,
producing banana-shaped arcs. Point
sources (such as stars and quasars,
which are either too small or too distant
to see in detail) typically remain points.
MULTIPLICATION Strong gravitational
lensing can produce multiple images.
Additional images always emerge in
pairs, and one of these images is mirrorinverted. Although the number of images
must be odd, one image is usually obscured, so observers see an even number.
SIMULATED DISTORTION demonstrates the lensing effects of a cluster of stars (top left).
Whenever a lens is not a single object but a collection of objects, the outcome can get
rather complicated. Astronomers visualize this by preparing a color map (top right) that
shows magnification as a function of position. The cluster magnifies a source of light to a
small (blue), moderate (green) or large (red) degree. The yellow lines are so-called
caustics, where the magnification is extremely high. The uneven magnification distorts a
perfectly respectable magazine logo (bottom left) into a phantasmagoria (bottom right).
SCIENTIFIC AMERICAN
NOVEMBER 2001
Copyright 2001 Scientific American, Inc.
JOACHIM WAMBSGANSS
■
1. How Lensing Works
Stars, galaxies or black holes can deflect light rays
from the straight and narrow
A
gravitational-lens system comprises four ingredients: a
distant source of light (star, galaxy or quasar), an intervening mass that acts as the lens (anything from a planet to a black hole), an observer on Earth, and the space in which
all three are embedded. The line that connects lens and observer
is called the optical axis.
Light always follows the shortest possible route between
two points. But Einstein showed that the shortest connection
between two points can be curved, just as the shortest path between two points on the surface of Earth is part of a great circle. As light rays approach the curved space near a cosmic body,
they will bend. The degree of deflection depends on how close
the rays get to the body and how massive this body is. The deflection angle is directly proportional to the mass and inversely proportional to the closest distance.
In many ways, gravitational lenses act like ordinary glass
lenses. One of the major differences is that ordinary lenses have
a well-defined focal point, whereas the gravitational varieties
produce focal lines or surfaces. The convex shape of an ordinary lens ensures that the deflection angle is directly proportional to the distance from the optical axis. All incoming parallel rays meet at the same point behind the lens—the focus. The
typical gravitational lens, however, causes light rays to experience smaller deflections the farther they are from the optical
axis. For this reason, parallel rays deflected by gravity meet at
different locations behind the lens, depending on how far away
from the optical axis they originate. Certain glass lenses have
the same effect; a good example is the bottom of a wineglass.
Another difference between gravitational lenses and ordinary glass lenses is that the former affect all wavelengths of light
equally. In other words, gravitational lensing is achromatic. For
glass lenses, the degree of deflection depends on the wavelength
of the light. Gravitational-lensing effects have been measured
throughout the electromagnetic spectrum, even in x-rays, which
cannot be focused by glass optics.
If the lens system is perfectly symmetric— source, lens and
observer are in alignment, and the lens is a point or sphere— the
rays converge somewhere along the optical axis and the resulting image is a ring (below). But if the system is asymmetric— if
the alignment is slightly off or the lens has an oblong mass distribution— the ring breaks up into discrete variegated images.
The lens magnifies different parts of the source by different
amounts. The highest magnification occurs for parts of the
source that happen to be on a curve known as the caustic. An
everyday example of caustics is the lacework of bright lines you
see on the bottom of a sunlit swimming pool; the ripples on the
surface of the water act as irregular lenses.
If the alignment is very far off or the lens mass distribution
is very spread out, the lensing is weak. Images are barely distorted or magnified. Although in this case the effects are difficult to discern for a single object, they can often be detected statistically by looking at large populations of objects.
CONVEX GLASS LENS
Light near the edge of a glass lens is
deflected more than light near the
optical axis. Thus, the lens focuses
parallel light rays onto a point.
GRAVITATIONAL LENS
SLIM FILMS
GRAVITATIONAL LENS SYSTEM
A gravitational lens (galaxy at center) takes light rays coming
from a distant galaxy and focuses some of them (purple cone) on
Earth. To observers, the light appears to have followed a straight
line (yellow cone), giving the illusion that it emanated from a ring.
www.sciam.com
Light near the edge of a gravitational
lens is deflected less than light near
the center. Thus, the lens focuses light
onto a line rather than a point.
SCIENTIFIC AMERICAN
Copyright 2001 Scientific American, Inc.
67
11. Quasars
As mighty as quasars are, they appear as mere dots
in most telescopes. Gravitational lensing can peer inside them
MULTIPLE QUASARS
EINSTEIN RINGS
Gravitational lensing became an observational science in 1979, when Dennis
Walsh of Jodrell Bank Observatory in England and his
colleagues discovered the
double quasar Q0957+561,
a pair of almost identical quasars right next to each other
in the sky. Today astronomers know of 64 double,
triple and multiple quasars
QUADRUPLE QUASAR Q2237+0305
separated by a few arcseconds or less. They are rare,
accounting for roughly one out of every 500 observed quasars.
The most comprehensive attempt so far to determine their
prevalence was the CLASS (Cosmic Lens All-Sky Survey) project, which mapped more than 10,000 radio sources and found
17 multiply imaged systems.
To make sure a grouping is an illusion rather than a real
cluster of quasars, observers go down a checklist: Do the
quasars lie at the same distance, as determined by measuring
the redshift? Are their spectra, which are as characteristic for
quasars as fingerprints are for humans, identical or at least very
similar? Is there a galaxy— a potential lens— between us and the
quasar? Finally, does the brightness of each quasar fluctuate
in exactly the same way?
The third of these criteria, the detection of a lens galaxy, is
not rigid, because it is possible that the galaxy is either very faint
or even completely dark. For instance, it may be a lump of gas
in which no stars have formed. The lens may not even be a galaxy
but rather an isolated black hole with a comparable mass. But
in every well-studied case of multiple-quasar images, astronomers
have been able to detect a more or less normal galaxy. One implication is that the universe does not contain large numbers of
dark galaxies or isolated supermassive black holes.
When a lens galaxy is spherically symmetric, it can redistribute the light of a background quasar or galaxy into
a complete circle. The diameter of the ring is propor- EINSTEIN RING B1938+666
tional to the square root of
the lensing mass— providing a very elegant way of determining the mass of the lens galaxy. About a dozen Einstein rings
are now known.
One of the most powerful applications of quasar lensing, first
suggested by Sjur Refsdal of Hamburg University in Germany
in 1964, is to gauge the Hubble constant, a measure of the size
and present expansion rate of the universe. Most other techniques to determine this value rely on a long ladder of distance
measurements, but the gravitational-lens method leaps to the
answer in a single bound.
When one of the images in a double quasar changes its
brightness, the other one usually does, too— but not at exactly
the same time. A delay is introduced by two effects: lensing asym-
Relative Brightness
Quasar Image 1
Dec 94
Jan 95
Feb 95
Mar 95
Apr 95
Quasar Image 2
Jan 96
Feb 96
Mar 96
Apr 96
May 96
Jun 96
Date
MATCHING BRIGHTNESS VARIATIONS of the double quasar Q0957+561
DOUBLE QUASAR HE1104–1805, straddling a faint lensing galaxy
68
metry (which forces the light rays that produce each image to
take paths of slightly different lengths) and the gravitational field
of the lens (which, according to relativity theory, reduces the
apparent velocity of light). From models of the shape and mass
distribution of the lens, astronomers can estimate the time delay
as a fraction of the total light-travel time. Then, by measuring
the time delay and dividing by this fraction— typically about
one ten-billionth— they can calculate the total light-travel time
from the quasar, hence its distance. Because the redshift measures the receding speed, the constant of proportionality between
SCIENTIFIC AMERICAN
NOVEMBER 2001
Copyright 2001 Scientific American, Inc.
ROBERT SCHMIDT University of Cambridge, SPACE TELESCOPE SCIENCE INSTITUTE AND NASA (quadruple quasar);
LINDSAY KING University of Bonn and Class Team (ring); SARA CHEN, SOURCE: TOMISLAV KUNDI ´Ć Renaissance Technologies (graph);
FREDERIC COURBIN Pontificia Universidad Catolica, Chile (double quasar)
HUBBLE CONSTANT
distance and velocity, the Hubble constant, can be calculated.
The technique was first applied to the double quasar
Q0957+561 (diagram on opposite page). Flickers in one of the
quasar images (blue) appear in the other (red) about 417 days
later, which implies that the quasar is about 14 billion lightyears away. Astronomers have now measured time delays from
seven multiple-quasar systems. The inferred value of the Hubble constant is lowish but matches those arrived at by other
techniques, within the error bars. The biggest uncertainty is the
complicated mass distribution in the lenses.
COSMOLOGICAL CONSTANT
Multiple quasars can also give insight into another infamous cosmological parameter, the cosmological constant. This constant,
or something like it, is needed to explain why the expansion of
the universe appears to be accelerating [see “The Quintessential
Universe,” by Jeremiah P. Ostriker and Paul J. Steinhardt; Scientific American, January]. The acceleration relates to lensing
because it makes the universe larger, which increases the probability that a quasar will be lensed. The more the expansion has
accelerated, the bigger the volume of space and the more likely
it is that an alignment between a galaxy and a distant quasar occurs (below). Therefore, the number of multiple quasars can put
an upper bound on the cosmological constant.
In 1998 Emilio E. Falco, Chris S. Kochanek and Jose A.
Muñoz of the Harvard-Smithsonian Center for Astrophysics
concluded that the cosmological constant cannot account for
more than 62 percent of the energy density of the universe. If
the constant were larger than this value, then observers should
find many more multiple quasars than they do. This analysis favors smaller values of the constant than do such cosmological
measurements as the brightness of distant supernovae, but the
difference is not statistically significant, and more recent studies have loosened the constraint a bit.
MICROLENSING OF QUASARS
Lensing is not always as obvious as in the examples above. If
a star does the lensing, for example, the images are so close together that even the best telescopes cannot resolve them. This
so-called microlensing effect is nonetheless measurable. Because the star is moving, the lens configuration— and therefore
the magnification— changes over time. If observers see a quasar
brighten and then dim in a particular way, they can infer that
a star passed in front and briefly magnified its image.
The problem is that quasars are unsteady; they tend to
brighten and dim on their own. To distinguish microlensing
fluctuations from the quasar’s intrinsic variability, astronomers
monitor multiple-quasar systems. If one of the images flickers
while the others do not, it may be because a star within the lens
galaxy has passed into the line of sight and temporarily added
an extra brightening to the effect already produced by the
galaxy as a whole. Intrinsic changes, on the other hand, will
show up in all the images. Since 1989 astronomers have confirmed microlensing in five multiple-quasar systems.
The brightness of the quasar increases smoothly until it hits
a caustic, and then it undergoes an abrupt drop. The effect depends on the size of the quasar: the smaller it is, the more abruptly the brightness varies. These patterns provide a way to measure the size of the quasar and probe its internal structure. The
brightness varies more abruptly in blue light than in red light.
Consequently, researchers conclude that the innermost parts of
a quasar are hotter and bluer than the outer parts. By monitoring caustic crossings using various color filters, astronomers can
reconstruct the brightness profile of the quasar.
a
MAGNIFICATION produced by star cluster: low (blue);
moderate (green); high (red); very high (yellow)
BRIGHTNESS FLUCTUATIONS
of a quasar as it passes behind
the star cluster
a
b
c
Relative Brightness
JOACHIM WAMBSGANSS (color map); SARA CHEN (graph)
APPARENT MOTION OF QUASAR
b
c
Time
www.sciam.com
SCIENTIFIC AMERICAN
Copyright 2001 Scientific American, Inc.
69
The lensing of galaxies betrays
the presence of dark matter
1v. Stars
The distortion of stars is too subtle to see directly
but shows up as a slow waxing and waning
GIANT LUMINOUS ARCS
MICROLENSING OF STARS
If the lens is not a single galaxy but an entire cluster of galaxies,
the image can be a kaleidoscope of strongly distorted arcs and
arclets. The first giant luminous arcs were discovered in 1986 independently by Roger Lynds of National Optical Astronomy
Observatory with Vahé Petrosian of Stanford University
and by Genevieve Soucail of
Midi-Pyrénées Observatory in
France and her colleagues. Almost 100 such arc clusters
have been identified so far,
one of the most dramatic being cluster Abell 2218 (left).
With the help of these images, astronomers can reconstruct the mass distribution inCLUSTER Abell 2218 distorts
side the cluster. The results,
images of more distant galaxies.
like those of other techniques
for measuring cluster masses,
imply that clusters are dominated by unseen dark matter. In addition, like multiple quasars, arcs can provide estimates of cosmological parameters such as the cosmological constant. In 1998
Matthias Bartelmann of the Max Planck Institute for Astrophysics in Garching, Germany, and his colleagues used the number of observed arc systems to measure the cosmological constant and came up with a lower value than have scientists using other methods. This discrepancy has not yet been resolved.
Lensing is an ideal way to ferret out the dark matter that lurks
in the outermost part of our galaxy, the halo. Some of this dark
matter may be exotic elementary particles, but some may comprise macroscopic objects that telescopes, for whatever reason,
cannot see directly: rogue planets, dead stars or black holes.
In 1986 Bohdan Paczyński of Princeton University suggested
a technique to search for such objects, collectively known as
MACHOs, or massive compact halo objects.
If a MACHO drifts in front of a background star, it will magnify that star and create a second image (below). Observers will
not be able to resolve the images, but they will notice a temporary brightening. The duration of the event is proportional to the
square root of the lens mass. This microlensing effect is relatively easy to distinguish from the other ways in which stars vary in
brightness. At any given moment, the chance of such an alignment is only about one in a million. But if observers monitor millions of stars at a time, they should occasionally see a microlensing event.
In the early 1990s several scientific teams, going by a slew
of contrived acronyms— notably the French EROS, the American-Australian MACHO and the Polish-American OGLE—
started to apply this method. Monitoring stars in the Large
Magellanic Cloud, a small satellite galaxy of the Milky Way,
the teams saw a total of almost two dozen microlensing events
over seven years. These events lasted from a few weeks to several months, implying that the objects had approximately half
COSMIC SHEAR
On extremely large scales, vaster even than galaxy clusters, agglomerations of matter tend to be too broad and smooth to act
as powerful lenses. Any distortion of galaxy images tends to get
lost in the natural variation of galaxy shapes. But when astronomers analyze thousands of galaxies, they can use statistical methods to look for tiny but systematic distortions. Last year
four teams— led by David J. Bacon of the University of Cambridge, Nick Kaiser of the
University of Hawaii Institute
for Astronomy, Ludovic van
Waerbeke of the Canadian Institute for Theoretical Astrophysics and David M. Wittman of Lucent Technologies in
Murray Hill, N.J.— independently discovered this very
weak lensing effect. The widespread shearing of galaxy images supports the view that the
RECONSTRUCTION of dark matter
universe is a giant cobweb of
distribution using weak lensing
matter interspersed with voids.
70
IF TELESCOPES HAD high enough resolution, a microlensing event would
look like this. In practice, observers see only that the star got brighter.
SCIENTIFIC AMERICAN
NOVEMBER 2001
Copyright 2001 Scientific American, Inc.
ANDREW FRUCHTER Space Telescope Science Institute and NASA (Abell 2218);
YANNICK MELLIER Institute of Astrophysics of Paris (dark matter); SARA CHEN (microlensing)
111. Galaxies
APPARENT MOTION OF
BACKGROUND STAR
Relative Brightness
BRIGHTNESS INCREASE
CAUSED BY STAR
BLIPS CAUSED
BY PLANET
SIMULATED STAR-PLANET
microlensing event shows how a
little planet can have a big effect on
brightness. The color maps (left)
show how magnification varies with
position; the three diagrams
correspond to three different
distances between planet and star.
As a background star moves
through one of these maps, it is
magnified to a small (blue),
moderate (green), high (red) or
very high (yellow) degree.
Consequently, the brightness
appears to fluctuate (right).
Time
the mass of the sun. The number of events, however, was too
low to explain more than a small fraction of the dark matter.
Analogous techniques for other galaxies suggest that their dark
matter cannot be made entirely of MACHOs, either.
The same teams also monitored stars toward the center of
our Milky Way and observed more than 500 microlensing
events in that direction, many more than expected. In this case,
the lenses were not MACHOs but most likely normal stars with
low mass. A small percentage appeared to be double stars,
which caused the brightness to vary abruptly because of caustic crossings. Monitoring such caustic crossings can reveal the
properties of stellar atmospheres and surfaces— the only way
that astronomers have been able to discern such fine detail on
distant stars. A few of the microlensing events may have been
caused by stellar-mass black holes.
JOACHIM WAMBSGANSS (color map); SARA CHEN (graph)
EXTRASOLAR PLANETS
Stellar microlensing can even detect planets. Several teams of
observers— the PLANET (Probing Lensing Anomalies Network) group led by Penny D. Sackett of the University of
Groningen in the Netherlands, the MPS (Microlensing Planet
Search) group headed by David P. Bennett of the University of
Notre Dame, and the MOA (Microlensing Observations in Astrophysics) group led by Philip Yock of Auckland University in
New Zealand— have taken a detailed look at some of the events
seen by the dark matter searches. In two cases, the observers
saw a blip— an extra burst of brightening that might have been
caused by a planet orbiting the lens star. Typically the blip lasted a few hours and boosted the brightness by a few percent.
Although these planet detections have not been independently confirmed, the principle is sound. It is only a question of
time until gravitational lensing reveals an entire list of con-
vincing planet candidates. Most other techniques look for the
planet’s effect on its parent star, which depends strongly on
planet mass or size. But with the lensing technique, even a lowmass planet produces a caustic that leads to high magnification
of the background star (above).
Five months ago a team of scientists headed by Kailash C.
Sahu of the Space Telescope Science Institute detected the flickering of a handful of stars in the central part of the Milky Way.
They tentatively interpreted it as microlensing by free-floating
planets in the globular cluster M22— an exciting claim that, if
confirmed, would have profound implications for the frequency of planetary-mass objects in the galaxy. Prior to this announcement, most astronomers had assumed that planets
would be found only orbiting a star, not off on their own in deep
space. It is yet another example of how scientists sometimes
come closest to the truth when they are studying “illusions.”
MORE TO E XPLORE
Gravitational Lenses. Peter Schneider, Jürgen Ehlers and Emilio E. Falco.
Springer, 1999.
Probing the Universe with Weak Lensing. Yannick Mellier in Annual
Review of Astronomy and Astrophysics, Vol. 37, pages 127–189; 1999.
Preprint available at www.arXiv.org/abs/astro-ph/9812172
Singularity Theory and Gravitational Lensing. Arlie O. Petters, Harold
Levine and Joachim Wambsganss. Birkhäuser, 2001.
Gravitational Lensing: A Brief Review. Joachim Wambsganss in Sources
and Scintillations: Refraction and Scattering in Radio Astronomy,
Proceedings of IAU Colloquium 182. Preprint available at
www.arXiv.org/abs/astro-ph/0012423
Gravitational Lensing in Astronomy. Joachim Wambsganss in
Living Reviews in Relativity. Available at
www.livingreviews.org/Articles/Volume1/1998-12wamb/
The Harvard-Smithsonian Center for Astrophysics/Arizona Space Telescope
Lens Survey: cfa-www.harvard.edu/glensdata/
University of Liège: vela.astro.ulg.ac.be/grav_lens/grav_lens.html
www.sciam.com
SCIENTIFIC AMERICAN
Copyright 2001 Scientific American, Inc.
71